Continuous Improvement: Insights from 'Topology' — LessWrong